Description
(previously the other half of #22 , split into its own issue)
String literals for OOV words
If the sentence contains words that are not in the lexicon, I would like to create those words as string literals. So "mimsy were the borogroves" would result in an otherwise normal GF tree, but with the subtrees StrA "mimsy"
and StrN "borogrove"
.
This feature should be optional: either command line arg, or check if the grammar contains StrA : String -> A
.
Future work: modify the PGF grammar?
The new majestic runtime will allow modifying PGFs on the fly. So when that is possible, I'd prefer to create proper lexicon entries mimsy_A
and borogrove_N
, using real GF smart paradigms, and insert them into the PGF.
(Similarly, the Backup* funs from #22 would also be possible to insert into the PGF.)
So once the new runtime is in place, I think that command line argument would be a better option. And if these features are added into gf-ud already before majestic runtime, it makes sense to just use command line arguments from the beginning.